Do not judge a kidney drug by the first three months of numbers

Design

Individual-patient meta-analysis of randomised trials

Population

66,449 participants in 49 CKD progression trials

Primary outcome

Association between early changes in GFR and albuminuria

Effect

Trial-level slope 0.00 (95% BCI −0.08 to 0.09); ACR −8.0% vs −6.6% per SD GFR decline

Early falls in GFR and albuminuria after starting a kidney-protective drug have been proposed as markers that it is working in that individual. This individual-patient meta-analysis tested that idea across 49 randomised trials with 66,449 participants, using GFR change to three months and albuminuria change to six months.

Across trials, there was no association between a treatment's effect on early GFR and its effect on albuminuria (slope 0.00). Within patients, a greater fall in GFR was linked to a fall in albumin-to-creatinine ratio, but the relationship was almost the same in the treatment arms (8.0% per standard deviation) and the control arms (6.6%). The pattern held across drug classes except immunosuppressants.

The implication is that much of what clinicians see in the first months reflects natural fluctuation and the disease itself rather than drug response. An early rise or fall in either measure, by itself, is a weak basis for deciding a drug is or is not working. Albuminuria remains a valid trial endpoint at the population level; this concerns reading it in one patient.

• Avoid stopping a kidney-protective drug solely because albuminuria has not fallen at three to six months.
• Do not assume a falling albuminuria in one patient proves the drug is working.
• Judge benefit by long-term eGFR trajectory and trial evidence for the drug class.
• Repeat albuminuria measurements, as single values vary considerably.
• Continue proven therapies such as RAS blockers and SGLT2 inhibitors unless there is a clear reason to stop.

Why it matters

Clinicians often stop or switch drugs based on early numbers that mostly reflect noise.

Don't overread it

This does not undermine albuminuria as a trial endpoint across populations.

The statistics, in plain English

A slope of 0.00 across trials means trials that changed GFR more did not change albuminuria more. Within patients, the GFR–albuminuria link was similar with and without active treatment, which points to natural variation rather than drug effect.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.